Re: [PATCH 0/2] gdb: resolve class name via DW_AT_signature in cooked index

[Simon Marchi <simon.marchi@efficios.com>]

On 9/24/26 12:59 AM, Simon Marchi wrote:> This is v5 of Andrew's fix for PR gdb/33447, where qualified member function
> lookup fails for C++ binaries built by Clang with -fdebug-types-section.  v4
> was here:
> 
>   https://inbox.sourceware.org/gdb-patches/e78a75720042dba3d3b052883d2d84d374ffc157.1789555833.git.aburgess@redhat.com/
> 
> I wrote this version addressing some comments I had on v4, with the hope that
> it can be ready to merge before the release planned on Friday.
> 
> The main changes are:
> 
>  - Patch 1 is new preparatory patch that does a bit of refactoring.  It splits
>    cooked_index_shard::finalize into separate steps, making it easier for patch
>    2 to add more steps.
> 
>  - The deferred names mechanism now uses a flag on the cooked_index_entries
>    (like the deferred parents)rather than a vector.  This is probably what Tom
>    meant in earlier reviews, but I did not understand at the time.  When I
>    finally understood how the deferred parent mechanism worked, it seemed
>    logical to implement the deferred names mechanism the same way, at least for
>    consistency.  Scanning over all index entries is relatively fast anyway, and
>    this step only runs if actually needed.
> 
> Andrew Burgess (1):
>   gdb: resolve class name via DW_AT_signature in cooked index
> 
> Simon Marchi (1):
>   gdb: split cooked index finalization into separate steps

Arf, it's too late here, I forgot the v5 tag :(.

Simon